Span-level PHI extraction on SHIELD HOSPITAL category 1.0 (test)
85PrecisionGemini Flash
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Gemini Flashmode=zero-shot extraction2026.05 | 85 | 82 | |
| GPT-OSS 120Bmode=zero-shot extraction2026.05 | 70 | 75 | |
| GPT-OSS 20Bmode=zero-shot extraction2026.05 | 68 | 71 | |
| Flash Litemode=zero-shot extraction2026.05 | 65 | 81 |